Efficient block-level space allocation for multi-version concurrency control data
Lockless allocation of a block from a page is initiated by computing a needed block size from an original block header value. Thereafter, a currently used block size and a capacity of the page are determined. A new block size is then computed based on the currently used block size and the needed block size and a remaining capacity of the page. A new block header value is subsequently generated that includes the determined new block size and the determined remaining capacity of the page The original block header value is then replaced with the new block header value. Related apparatus, systems, techniques and articles are also described.
Latest SAP SE Patents:
The subject matter described herein relates an in-memory database system providing efficient block-level space allocation for multi-version concurrency control data.
BACKGROUNDIn-memory databases are database management systems in which data is primarily stored transiently; namely in main memory. In order to obtain optimum performance, as much data as possible must be kept in memory. However, given the large number of concurrent transactions, multi-version concurrency control (MVCC) mechanisms are adopted to provide point in time consistent views on the underlying data.
SUMMARYIn one aspect, lockless allocation of a block from a page is initiated by computing a needed block size from an original block header value. Thereafter, a currently used block size and a capacity of the page are determined. A new block size is then computed based on the currently used block size and the needed block size and a remaining capacity of the page. A new block header value is subsequently generated that includes the determined new block size and the determined remaining capacity of the page The original block header value is then replaced with the new block header value.
The replacing can use a compare and swap operation. The determining, computing, and generating can be iteratively implemented until the original block header value is replaced with the new block header value.
A pointer to the block can be returned after replacing the original block header value with the new block header value. The pointer can point to a location in the block equal to the starting of the page plus the currently used block size. The location can be part of a block handle.
In another interrelated aspect, allocation of a block from a new page is initiated in a multi-version concurrency control (MVCC) page chain by obtaining a first current last page number in the MVCC page chain. Thereafter, a mutex lock is acquired that prevents allocation of pages in the page chain. Next, a second current last page number is obtained in the MVCC page chain. Subsequently, it is attempted to allocate a block from a last page in the MVCC page chain if the first current last page number and the second page number differ. A new MVCC page can be allocated in the MVCC page chain if the attempt to allocate the block fails or if there is no change in the last page. A new block can be allocated with a specified block size from a newly allocated page using lockless block allocation logic that returns a pointer to the block. The allocated new page is then appended to the MVCC page chain. The mutex lock is then released and a block handle holding the allocated new block is then returned.
In still a further interrelated aspect, allocation of a block from a last page in a multi-version concurrency control (MVCC) page chain is initiated by attempting to return a block from a lockless free list (e.g, a free reserved stub list, a free stub list, or free MVCC blocks, etc.) implemented using a free queue, or, if no block is available, then allocating a block from the last page. Thereafter, a last page of an MVCC page chain is loaded. Thereafter, it is attempted to allocate the block using lockless block allocation from the last page in the MVCC page chain if the last page is valid. The block is allocated from a new page if the attempt to allocate fails or there are no changes in the pages. In addition, a block handle associated with allocated new block is returned.
The free list can include destroyed MVCC blocks and can be managed as a lock-less queue.
The block can be returned from an MVCC free block list for blocks with header and data sections. The block can be returned from a free reserved stub list or free stub list for stub timestamp blocks (with the block only having a header section, i.e., no data). The block can be returned from a free stub block list the block from the pages using a lock-less block allocation algorithm (with the block only having a header section, i.e., no data). Allocation of a block can be attempted a block from the MVCC pages of the MVCC page chain either from a last page or from a newly allocated page.
Non-transitory computer program products (i.e., physically embodied computer program products) are also described that store instructions, which when executed by one or more data processors of one or more computing systems, causes at least one data processor to perform operations herein. Similarly, computer systems are also described that may include one or more data processors and memory coupled to the one or more data processors. The memory may temporarily or permanently store instructions that cause at least one processor to perform one or more of the operations described herein. In addition, methods can be implemented by one or more data processors either within a single computing system or distributed among two or more computing systems. Such computing systems can be connected and can exchange data and/or commands or other instructions or the like via one or more connections, including but not limited to a connection over a network (e.g. the Internet, a wireless wide area network, a local area network, a wide area network, a wired network, or the like), via a direct connection between one or more of the multiple computing systems, etc. Further, the current subject matter can be implemented and/or form part of an in-memory database such as columnar in-memory database.
The subject matter described herein provides many technical advantages. For example, the current subject matter is advantageous in that it expands on available row state information to provide lockless allocation of blocks of memory from a page. In particular, the current subject matter enables for differentiation among blocks that have been cleaned up and blocks that have been garbage collected.
The details of one or more variations of the subject matter described herein are set forth in the accompanying drawings and the description below. Other features and advantages of the subject matter described herein will be apparent from the description and drawings, and from the claims.
Like reference symbols in the various drawings indicate like elements.
DETAILED DESCRIPTIONThe current subject matter includes a number of aspects that can be applied individually or in combinations of one or more such aspects to support a unified database table approach that integrates the performance advantages of in-memory database approaches with the reduced storage costs of on-disk database approaches. The current subject matter can be implemented in database systems using in-memory OLAP, for example including databases sized at several terabytes (or more), tables with billions (or more) of rows, and the like; systems using in-memory OLTP (e.g. enterprise resource planning or ERP system or the like), for example in databases sized at several terabytes (or more) with high transactional volumes; and systems using on-disk OLAP (e.g. “big data,” analytics servers for advanced analytics, data warehousing, business intelligence environments, or the like), for example databases sized at several petabytes or even more, tables with up to trillions of rows, and the like.
The current subject matter can be implemented as a core software platform of an enterprise resource planning (ERP) system, other business software architecture, or other data-intensive computing application or software architecture that runs on one or more processors that are under the control of a specific organization. This arrangement can be very effective for a large-scale organization that has very sophisticated in-house information technology (IT) staff and for whom a sizable capital investment in computing hardware and consulting services required to customize a commercially available business software solution to work with organization-specific business processes and functions is feasible.
A database management agent 160 or other comparable functionality can access a database management system 170 that stores and provides access to data (e.g. definitions of business scenarios, business processes, and one or more business configurations as well as data, metadata, master data, etc.) relating to definitions of the business scenarios, business processes, and one or more business configurations, and/or concrete instances of data objects and/or business objects that are relevant to a specific instance of a business scenario or a business process, and the like. The database management system 170 can include at least one table 180 and additionally include parallelization features consistent with those described herein.
To achieve a best possible compression and also to support very large data tables, a main part of the table can be divided into one or more fragments.
Fragments 330 can advantageously be sufficiently large to gain maximum performance due to optimized compression of the fragment and high in-memory performance of aggregations and scans. Conversely, such fragments can be sufficiently small to load a largest column of any given fragment into memory and to sort the fragment in-memory. Fragments can also be sufficiently small to be able to coalesce two or more partially empty fragments into a smaller number of fragments. As an illustrative and non-limiting example of this aspect, a fragment can contain one billion rows with a maximum of 100 GB of data per column. Other fragment sizes are also within the scope of the current subject matter. A fragment can optionally include a chain of pages. In some implementations, a column can also include a chain of pages. Column data can be compressed, for example using a dictionary and/or any other compression method. Table fragments can be materialized in-memory in contiguous address spaces for maximum performance. All fragments of the database can be stored on-disk, and access to these fragments can be made based on an analysis of the data access requirement of a query. Each fragment will have its own set of MVCC pages which are hosting the MVCC blocks that stores the (row state or timestamp information) metadata used for determining visibility of the rows. MVCC data is the metadata stored along with the actual columns data for determining the visibility of the row used along with the read snapshot timestamp. A transaction manger 230 can be coupled to the main store 210 (and the corresponding MVCC store 211) as well as the delta store 206 (and the corresponding MVCC store 207) to provide snapshot timestamps for read operations along with MVCC data (which in turn provides visibility data for the corresponding row).
Referring again to
Also as shown in
A single RowID space can be used across pages in a page chain. A RowID, which generally refers to a logical row in the database, can be used to refer to a logical row in an in-memory portion of the database and also to a physical row in an on-disk portion of the database. A row index typically refers to physical 0-based index of rows in the table. A 0-based index can be used to physically address rows in a contiguous array, where logical RowIDs represent logical order, not physical location of the rows. In some in-memory database systems, a physical identifier for a data record position can be referred to as a UDIV or DocID or RowPos. Distinct from a logical RowID, the UDIV or DocID or RowPos (or a comparable parameter) can indicate a physical position of a row (e.g. a data record), whereas the RowID indicates a logical position. To allow a partition of a table to have a single RowID and row index space consistent with implementations of the current subject matter, a RowID can be assigned a monotonically increasing ID for newly-inserted records and for new versions of updated records across fragments. In other words, updating a record will change its RowID, for example, because an update is effectively a deletion of an old record (having a RowID) and insertion of a new record (having a new RowID). Using this approach, a delta store of a table can be sorted by RowID, which can be used for optimizations of access paths. Separate physical table entities can be stored per partition, and these separate physical table entities can be joined on a query level into a logical table.
Functional block diagram 500 also illustrates a read operation 520. Generally, read operations can have access to all fragments (i.e., active fragment 512 and closed fragments 516). Read operations can be optimized by loading only the fragments that contain data from a particular query. Fragments that do not contain such data can be excluded. In order to make this decision, container-level metadata (e.g., a minimum value and/or a maximum value) can be stored for each fragment. This metadata can be compared to the query to determine whether a fragment contains the requested data.
With reference to diagram 600 of
With reference to diagram 800 of
MVCC data can be persisted and can be stored in MVCC Pages (which can, for example, be derived from a regular block page but with an MVCC tag) and which can, for example, have a size of 256K. MVCC data can be addressed by indexing rowPos into a blocks in the MVCC page. Date can then be stored in blocks. These blocks can have varying sizes. In one example, the block can be 8K plus the size of the MVCC block header. Timestamp blocks can include CTS and DTS. In one example, the timestamp blocks can holds 1K entries, 8 byte value. RowState blocks can, for example, hold 32K entries with 2 bits per row. Stub blocks can include STUB_CTS, STUB_DTS, and STUB_RESERVED and can, for example, be 32 bytes and they can act as place holders from the garbage collected range (i.e. the range of rows are fully visible/invisible).
Blocks can be used in order to perform atomic operations and store data that can be indexed using bit-wise operations. For example, 8K data can easily be addressed by shifting (10 bits for TS or 15 bits for RS) of rowPos. Further, Bit shift operations are more efficient than multiplication, division or modulo operations.
MVCC blocks can have an index structure that can map rowPos to block. Such an index structure can provide lockless read and write operations. Further, the index structure can be dynamically grown so that it can increase the size of the structure while allocating new blocks. In addition, the MVCC index structure is advantageous in that it allows for consistent reads while, at the same time, dynamically growing the structure. The index structure can use versioned vectors [Header, Versioned data object] with additional logic in APIs that can store the MVCC information. This index structure can be used to retrieve the MVCC block information for the given RowPos. At the same time, the writes to the index structure can be synchronized using atomic operations. As the block size is fixed to 8K, block size along with size of the MVCC data (8 byte for timestamp and 2 bits row state) can be used to compute the offset into the index structure for retrieving the block.
The MVCC block can be represented by minimal information that can be stored using atomic operations. For example, the MVCC block header can be 32 bytes which cannot be stored using atomic APIs. Therefore, a structure referred to herein as a block handle (or sometimes a BlockHandle) can be stored in the versioned vector. The block handle can include a pointer to a corresponding block and a handle to a page associated with the block. In addition, the row position (rowPos) can be used to determine the location of the MVCC cell within an MVCC block.
To effect such functionality, the MVCC object 920 can include a row state vector 922, a confirm timestamp vector 924, and a delete timestamp vector 926. In addition, the MVCC block handler 910 can include a page chain 930 of pages which can each include an MVCC page header 932. The page header 932 can, in turn, specify/encapsulate a timestamp block 934, a reserved stub block 936, and a row state block 938. The MVCC object 920 (as part of the insert operation) can request a new MVCC block from the MVCC block handler 910 if so indicated by the row state vector 922. Thereafter, a new MVCC block is allocated and the row state block 938 in the page chain 930 is updated to reflect the same. Further, after the insert operation commits, the commit timestamp is stored in the timestamp block 934 and the CTS vector 924 is accordingly updated. The reserved stub block 936 can be generated as part of the timestamp block 934 allocation/generation and can be used as described below.
As referenced above, a stub block (referred to herein sometimes just as a stub) can represent a range of row position values having timestamp values that have been garbage collected. During cleanup/garbage collection, if the timestamp values of all the rows/entries in a block are zeros, then such block can be replaced with a stub. After replacing the stub, the earlier timestamp block can be moved to a free list and will be later used by another task.
Stub blocks can persist the fact that rows associated with the range are committed deleted rows (INVISIBLE, 0, 0) rather than unused rowPos values (INVISIBLE, INT64_MAX, INT64_MAX). As a result, stub blocks can extend the values provided by the row state bits. This differentiation is required to generate a list of valid values and invalid values.
Stub blocks can be allocated while allocating the timestamp block and can be used by a cleanup thread for replacing it with a regular block without allocating (which can lead to resource related exceptions) from a page. Further, stub blocks allocated as part of timestamp block allocation can be characterized as reserved stub blocks which can be maintained in a free reserved stub block list. Cleanup threads use one of the reserved stub blocks from the list and marks it as STUB_CTS or STUB_DTS based on the type block information vector.
With the deletion operation, a client seeks to delete an entry at row 1800 of the unified table fragment 1510. Thereafter, an MVCC object 1520 (not necessarily the same object as used in the insertion operation), is accessed which, in turn, points to a first block 1530 (forming part of the MVCC page 1760). The first block 1530 comprises the row state vector 1535 which characterizes the particular state for the row (which in this case shows the row as visible). After the entry at 1800 is deleted, the DTS vector 1555 in a linked block 1550 is updated with the corresponding timestamp.
After this sequence of operations, the MVCC page 1560 comprises a first slot 1562 comprising the row state from the first block 1530, a second slot 1564 comprising a reserved stub for the CTS timestamp, a third slot 1566 comprising a CTS timestamp from the second block 1540, a fourth slot 1568 comprising a reserved stub for for the DTS timestamp, and a fifth slot 1570 comprising a DTS timestamp.
After the state in diagram 1500 of
One or more aspects or features of the subject matter described herein can be realized in digital electronic circuitry, integrated circuitry, specially designed application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs) computer hardware, firmware, software, and/or combinations thereof. These various aspects or features can include implementation in one or more computer programs that are executable and/or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device. The programmable system or computing system may include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
These computer programs, which can also be referred to as programs, software, software applications, applications, components, or code, include machine instructions for a programmable processor, and can be implemented in a high-level procedural language, an object-oriented programming language, a functional programming language, a logical programming language, and/or in assembly/machine language. As used herein, the term “machine-readable medium” refers to any computer program product, apparatus and/or device, such as for example magnetic discs, optical disks, memory, and Programmable Logic Devices (PLDs), used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term “machine-readable signal” refers to any signal used to provide machine instructions and/or data to a programmable processor. The machine-readable medium can store such machine instructions non-transitorily, such as for example as would a non-transient solid-state memory or a magnetic hard drive or any equivalent storage medium. The machine-readable medium can alternatively or additionally store such machine instructions in a transient manner, such as for example as would a processor cache or other random access memory associated with one or more physical processor cores.
To provide for interaction with a user, one or more aspects or features of the subject matter described herein can be implemented on a computer having a display device, such as for example a cathode ray tube (CRT) or a liquid crystal display (LCD) or a light emitting diode (LED) monitor for displaying information to the user and a keyboard and a pointing device, such as for example a mouse or a trackball, by which the user may provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well. For example, feedback provided to the user can be any form of sensory feedback, such as for example visual feedback, auditory feedback, or tactile feedback; and input from the user may be received in any form, including, but not limited to, acoustic, speech, or tactile input. Other possible input devices include, but are not limited to, touch screens or other touch-sensitive devices such as single or multi-point resistive or capacitive trackpads, voice recognition hardware and software, optical scanners, optical pointers, digital image capture devices and associated interpretation software, and the like.
In the descriptions above and in the claims, phrases such as “at least one of” or “one or more of” may occur followed by a conjunctive list of elements or features. The term “and/or” may also occur in a list of two or more elements or features. Unless otherwise implicitly or explicitly contradicted by the context in which it is used, such a phrase is intended to mean any of the listed elements or features individually or any of the recited elements or features in combination with any of the other recited elements or features. For example, the phrases “at least one of A and B;” “one or more of A and B;” and “A and/or B” are each intended to mean “A alone, B alone, or A and B together.” A similar interpretation is also intended for lists including three or more items. For example, the phrases “at least one of A, B, and C;” “one or more of A, B, and C;” and “A, B, and/or C” are each intended to mean “A alone, B alone, C alone, A and B together, A and C together, B and C together, or A and B and C together.” In addition, use of the term “based on,” above and in the claims is intended to mean, “based at least in part on,” such that an unrecited feature or element is also permissible.
The subject matter described herein can be embodied in systems, apparatus, methods, and/or articles depending on the desired configuration. The implementations set forth in the foregoing description do not represent all implementations consistent with the subject matter described herein. Instead, they are merely some examples consistent with aspects related to the described subject matter. Although a few variations have been described in detail above, other modifications or additions are possible. In particular, further features and/or variations can be provided in addition to those set forth herein. For example, the implementations described above can be directed to various combinations and subcombinations of the disclosed features and/or combinations and subcombinations of several further features disclosed above. In addition, the logic flows depicted in the accompanying figures and/or described herein do not necessarily require the particular order shown, or sequential order, to achieve desirable results. Other implementations may be within the scope of the following claims.
Claims
1. A method comprising:
- initiating lockless allocation of a block from a page by computing a needed block size from an original value in a page header, the original value comprising a currently used block size of the page and a capacity of the page;
- computing a next used block size based on the currently used block size and the needed block size and a remaining capacity of the page after allocation of the block from the page;
- generating a new value comprising the next used block size of the page and the remaining capacity of the page; and
- replacing the original value in the page header with the new value.
2. The method of claim 1, wherein the replacing uses a compare and swap operation.
3. The method of claim 2, wherein the computing and generating are iteratively implemented until the original value is replaced with the new value in the page header.
4. The method of claim 1 further comprising:
- returning a pointer to the block after replacing the original value in the page header with the new value in the page header.
5. The method of claim 4, wherein the pointer points to a location in the page that is computed from the block size from the page header plus the currently used block size.
6. The method of claim 4, wherein the location is part of a block handle.
7. The method of claim 1, wherein the initiating computing, generating, and replacing are implemented by at least one hardware data processor.
8. The method of claim 7, wherein the at least one hardware data processor forms part of an in-memory database.
9. A method comprising:
- initiating allocation of a block from a new page in a multi-version concurrency control (MVCC) page chain by obtaining a first current last page number in the MVCC page chain;
- acquiring a mutex lock that prevents allocation of pages in the page chain;
- obtaining a second current last page number in the MVCC page chain;
- attempting to allocate a block from a last page in the MVCC page chain if the first current last page number and the second current last page number differs;
- allocating a new MVCC page in the MVCC page chain if the attempt to allocate the block fails or if there is no change in the last page;
- allocating a new block with a specified block size from a newly allocated page using lockless block allocation logic that returns a pointer to the block;
- appending the allocated new page to the MVCC page chain; and
- releasing the mutex lock and returning a block handle holding the allocated new block.
10. The method of claim 9, wherein the method is implemented by at least one hardware data processor.
11. The method of claim 10, wherein the at least one hardware data processor forms part of an in-memory database.
12. A method comprising:
- initiating allocation of a block from a last page in a multi-version concurrency control (MVCC) page chain by attempting to return a block from a lockless free list, or, if no block is available, then allocating a block from the last page;
- loading a last page of the MVCC page chain;
- attempting to allocate the block using lockless block allocation from the last page in the MVCC page chain if the last page is valid;
- allocating the block from a new page if the attempt to allocate fails or there are no changes in the pages; and
- returning a block handle associated with allocated new block.
13. The method of claim 12, wherein the lockless free list is a free reserved stub list, a free stub list, or free MVCC block list.
14. The method of claim 13, wherein each of the operations are implemented by at least one hardware data processor.
15. The method of claim 14, wherein the at least one hardware data processor forms part of an in-memory database.
16. The method of claim 13, wherein the free list comprises destroyed MVCC blocks and is managed as a lock-less queue.
17. The method of claim 13 further comprising:
- returning the block from an MVCC free block list for blocks with header and data sections.
18. The method of claim 13 further comprising:
- returning the block from a free reserved stub list or free stub list for stub timestamp blocks;
- wherein the block comprises only a header section.
19. The method of claim 13 further comprising:
- returning the block from a free stub block list or allocating the block from the pages using a lock-less block allocation algorithm;
- wherein the block comprises only a header section.
20. The method of claim 13 further comprising:
- attempting to allocate a block from the MVCC pages of the MVCC page chain either from a last page or from a newly allocated page.
5222235 | June 22, 1993 | Hintz et al. |
5594898 | January 14, 1997 | Dalal et al. |
5701480 | December 23, 1997 | Raz |
5717919 | February 10, 1998 | Kodavalla |
5758145 | May 26, 1998 | Bhargava et al. |
5794229 | August 11, 1998 | French et al. |
5870758 | February 9, 1999 | Bamford et al. |
5933833 | August 3, 1999 | Musashi |
6275830 | August 14, 2001 | Muthukkaruppan et al. |
6282605 | August 28, 2001 | Moore |
6490670 | December 3, 2002 | Collins |
6606617 | August 12, 2003 | Bonner et al. |
6668263 | December 23, 2003 | Cranston et al. |
6754653 | June 22, 2004 | Bonner et al. |
6865577 | March 8, 2005 | Sereda |
7698712 | April 13, 2010 | Schreter |
8024296 | September 20, 2011 | Gopinathan et al. |
8161024 | April 17, 2012 | Renkes et al. |
8364648 | January 29, 2013 | Sim-Tang |
8510344 | August 13, 2013 | Briggs et al. |
8650583 | February 11, 2014 | Schreter |
8732139 | May 20, 2014 | Schreter |
8768891 | July 1, 2014 | Schreter |
9058268 | June 16, 2015 | Ostiguy et al. |
9098522 | August 4, 2015 | Lee et al. |
9141435 | September 22, 2015 | Wein |
9262330 | February 16, 2016 | Muthukumarasamy |
9268810 | February 23, 2016 | Andrei et al. |
9275095 | March 1, 2016 | Bhattacharjee et al. |
9275097 | March 1, 2016 | DeLaFranier et al. |
9305046 | April 5, 2016 | Bhattacharjee et al. |
9372743 | June 21, 2016 | Sethi et al. |
9489409 | November 8, 2016 | Sharique et al. |
20010051944 | December 13, 2001 | Lim et al. |
20020107837 | August 8, 2002 | Osborne et al. |
20020156798 | October 24, 2002 | Larue et al. |
20030028551 | February 6, 2003 | Sutherland |
20030065652 | April 3, 2003 | Spacey |
20030204534 | October 30, 2003 | Hopeman et al. |
20030217075 | November 20, 2003 | Nakano et al. |
20040034616 | February 19, 2004 | Witkowski et al. |
20040054644 | March 18, 2004 | Ganesh et al. |
20040249838 | December 9, 2004 | Hinshaw et al. |
20050027692 | February 3, 2005 | Shyam et al. |
20050097266 | May 5, 2005 | Factor |
20050234868 | October 20, 2005 | Terek et al. |
20080046444 | February 21, 2008 | Fachan et al. |
20080247729 | October 9, 2008 | Park |
20090064160 | March 5, 2009 | Larson et al. |
20090094236 | April 9, 2009 | Renkes et al. |
20090254532 | October 8, 2009 | Yang et al. |
20090287737 | November 19, 2009 | Hammerly |
20100008309 | January 14, 2010 | Cheng et al. |
20100082545 | April 1, 2010 | Bhattacharjee et al. |
20100287143 | November 11, 2010 | Di Carlo |
20110087854 | April 14, 2011 | Rushworth |
20110145835 | June 16, 2011 | Rodrigues |
20110153566 | June 23, 2011 | Larson et al. |
20110302143 | December 8, 2011 | Lomet |
20120011106 | January 12, 2012 | Reid et al. |
20120047126 | February 23, 2012 | Branscome |
20120102006 | April 26, 2012 | Larson et al. |
20120137081 | May 31, 2012 | Shea |
20120179877 | July 12, 2012 | Shriraman et al. |
20120191696 | July 26, 2012 | Renkes et al. |
20120233438 | September 13, 2012 | Bak et al. |
20120265728 | October 18, 2012 | Plattner et al. |
20120284228 | November 8, 2012 | Ghosh et al. |
20130054936 | February 28, 2013 | Davis |
20130091162 | April 11, 2013 | Lewak |
20130097135 | April 18, 2013 | Goldberg |
20130346378 | December 26, 2013 | Tsirogiannis |
20140025651 | January 23, 2014 | Schreter |
20140101093 | April 10, 2014 | Lanphear et al. |
20140214334 | July 31, 2014 | Plattner et al. |
20140279930 | September 18, 2014 | Gupta |
20140279961 | September 18, 2014 | Schreter et al. |
20150039573 | February 5, 2015 | Bhattacharjee et al. |
20150089125 | March 26, 2015 | Mukherjee et al. |
20150113026 | April 23, 2015 | Sharique et al. |
20150142819 | May 21, 2015 | Florendo |
20160103860 | April 14, 2016 | Bhattacharjee et al. |
20160125022 | May 5, 2016 | Rider et al. |
20160147445 | May 26, 2016 | Schreter et al. |
20160147447 | May 26, 2016 | Blanco et al. |
20160147448 | May 26, 2016 | Schreter et al. |
20160147449 | May 26, 2016 | Andrei et al. |
20160147457 | May 26, 2016 | Legler et al. |
20160147459 | May 26, 2016 | Wein et al. |
20160147617 | May 26, 2016 | Lee et al. |
20160147618 | May 26, 2016 | Lee et al. |
20160147750 | May 26, 2016 | Blanco et al. |
20160147776 | May 26, 2016 | Florendo et al. |
20160147778 | May 26, 2016 | Schreter et al. |
20160147786 | May 26, 2016 | Andrei et al. |
20160147801 | May 26, 2016 | Wein et al. |
20160147804 | May 26, 2016 | Wein et al. |
20160147806 | May 26, 2016 | Blanco et al. |
20160147808 | May 26, 2016 | Schreter et al. |
20160147809 | May 26, 2016 | Schreter et al. |
20160147811 | May 26, 2016 | Eluri et al. |
20160147812 | May 26, 2016 | Andrei et al. |
20160147813 | May 26, 2016 | Lee et al. |
20160147814 | May 26, 2016 | Goel et al. |
20160147819 | May 26, 2016 | Schreter et al. |
20160147820 | May 26, 2016 | Schreter |
20160147821 | May 26, 2016 | Schreter et al. |
20160147834 | May 26, 2016 | Lee et al. |
20160147858 | May 26, 2016 | Lee et al. |
20160147859 | May 26, 2016 | Lee et al. |
20160147861 | May 26, 2016 | Schreter et al. |
20160147862 | May 26, 2016 | Schreter et al. |
20160147904 | May 26, 2016 | Wein et al. |
20160147906 | May 26, 2016 | Schreter et al. |
2778961 | September 2014 | EP |
WO-01/29690 | April 2001 | WO |
- “HANA Database Lectures” Mar. 2014 Yeditepe Üniversitesi Bilgisayar Mühendisli{hacek over (g)}i Bölümü. Web. Apr. 22, 2016. <http://cse.yeditepe.edu.tr/˜odemir/spring2014/cse415/HanaDatabase.pdf>.
- “HANA Persistence: Shadow Pages.” Jun. 2013. Yeditepe Üniversitesi Bilgisayar Mühendisli?i Bölümü. Web. Apr. 21, 2016; <http://cse.yeditepe.edu.tr/˜odemir/spring2014/cse415/Persistency.pptx>.
- “Optimistic concurrency control.” Wikipedia: The Free Encyclopedia. Wikimedia Foundation, Inc., Jul. 19, 2014. Web. Mar. 3, 2016.
- Brown, E. et al. “Fast Incremental Indexing for Full-Text Information Retrieval.” VLDB '94 Proceedings of the 20th International Conference on Very Large Data Bases. San Francisco: Morgan Kaufmann, 1994.
- Jens Krueger et al. “Main Memory Databases for Enterprise Applications.” Industrial Engineering and Engineering Management (IE&EM), 2011 IEEE 18th International Conference on, IEEE, Sep. 3, 2011, pp. 547-557, XP032056073.
- Lemke, Christian, et al. “Speeding Up Queries in Column Stores.” Data Warehousing and Knowledge Discovery Lecture Notes in Computer Science (2010): 117-29. Web. Apr. 21, 2016.
- Lu, Andy. “SAP HANA Concurrency Control.” SAP Community Network. Oct. 28, 2014. Web. Apr. 22, 2016. <http://scn.sap.com/docs/DOC-57101>.
- Mumy, Mark. “SAP Sybase IQ 16.0 Hardware Sizing Guide.” SAP Community Network. May 12, 2013. Web. Apr. 21, 2016. <http://www.sdn.sap.com/irj/scn/go/portal/prtroot/docs/library/uuid/c0836b4f-429d-3010-a686-c35c73674180?QuickLink=index&overridelayout=true&58385785468058>.
- “NBit Dictionary Compression,” Sybase, May 23, 2013. Web. Mar. 15, 2017 <http://infocenter.sybase.com/help/index.jsp?topic=/com.sybase.infocenter.dc1777.1600/doc/h tml/wi11345808527844.html>.
Type: Grant
Filed: Nov 25, 2014
Date of Patent: Jan 23, 2018
Patent Publication Number: 20160147448
Assignee: SAP SE (Walldorf)
Inventors: Ivan Schreter (Malsch), Amarnadh Sai Eluri (Pleasanton, CA)
Primary Examiner: Ryan Bertram
Assistant Examiner: Trang Ta
Application Number: 14/553,680
International Classification: G06F 17/30 (20060101); G06F 3/06 (20060101); G06F 12/00 (20060101);